Because an amino acid can be encoded by as many as six codons, there are many different ways to encode any particular protein. Biases exist in the way codons are used. One well- known bias is the codon bias, which is simply that some codons are used more than others. Less well-known is that there is also a "codon pair bias", such that some codons "prefer" to be adjacent to certain other codons. This codon pair bias is completely separate and independent from the codon bias. Recently, we have found that when viruses are re-coded to have a bad codon pair bias, the viruses are attenuated, in extreme cases to inviability. It appears that attenuation of a virus via a bad codon pair bias can be used to make a live, attenuated vaccine. However, while the procedure works, nothing whatever is known about the mechanim by which codon pair bias causes attenuation, and this lack of knowledge is slowing the work with viral vaccines. In this proposal, we will investigate, for the first time, the mechanism of attenuation by bad codon pair bias. This will be done in yeast, where we have recently shown there are strong codon pair bias effects. We will confirm the effects of codon pair bias in yeast using two codon pair de-optimized synthetic yeast genes, dHIS3 and dLYS2. We will test the idea that bad codon pair bias slows translation using polysome profiling and ribosome density mapping. We will test the idea that bad codon pair bias causes inaccurate translation and protein degradation by turning off proteasomal degradation, and also using other methods. We have already selected yeast mutants that are apparently less sensitive than wild-type to codon pair bias, and we will characterize these mutants and identify the mutant genes. Finally we will do a high- throughput study of (initially) 22,000 different encodings of HIS3 to correlate particular encodings with the strength of gene function. Our long term goal is to understand the mechanism of attenuation by bad codon pair bias to facilitate the development of live, attenuated viral vaccines, and also to facilitate the tuning of gene expression. PUBLIC HEALTH RELEVANCE: One way to make an anti-viral vaccine is to mutate the virus to weaken it, then use this weakened virus as a vaccine (e.g., FluMist, a live Flu vaccine). However there are many difficulties in doing this. Recently we have found a new method, codon pair de-optimization, for this purpose. Although the method works, we do not know why it works. Here, we will study the mechanism of attenuation by codon pair de-optimization.